Beschreibung
therefore attempted to achieve the aims while avoiding these problems in two ways. The first is the separation into the theoretical discussion of Part 1 and the applied examples of Part 2. The textual examples in Part 1 are all of an abstract form and the real examples of Part 2 show what happens with real data. Following the Contents is a table showing the types of analysis used in each of the studies in Part 2. The table is designed to act as a link to real examples of each sub-type of analysis. Thus one can either read the Part 2 examples in their entirety, as examples of complete analyses, or one can use the above-mentioned of analysis. table to key into several different examples of a particular type The second way we have attempted to satisfy both aims is by inserting simple numerical examples into Part 1, set up as inset passages so as not to interrupt the theoretical development. They may thus either be read in parallel with the theory or skipped to be returned to later. The difference between these examples and those of Part 2, apart from one of simplicity, is that the Part 1 examples involve artificial data. We do not attempt complete analyses in these examples, simply using the data to illustrate isolated points.
Inhalt
1 The Theory.- 1 Introduction.- 1.1 What it is about.- 1.2 Before we start.- 1.3 Further reading.- 2 Between-subjects analysis.- 2.1 Introduction.- 2.2 One-factor designs.- 2.3 Comparisons on a single factor.- 2.4 Estimating the sizes of patterns of differences.- 2.5 A special case.- 2.6 Testing single contrasts.- 2.7 Testing comparisons.- 2.8 Random effects models.- 2.9 The overall mean level.- 2.10 Multi-factor designs.- 2.11 Contrasts in multi-factor designs.- 2.12 Estimation in multi-factor designs.- 2.13 Testing in multi-factor designs.- 2.14 Nested designs.- 2.15 Some important comparisons.- 2.16 Further reading.- 3 Within-subjects analysis.- 3.1 Introduction.- 3.2 Types of question.- 3.3 Multiple univariate studies.- 3.4 Structured multivariate studies: a simple example.- 3.5 Repeated-measures structures: trend analysis.- 3.6 Repeated measures by univariate methods.- 3.7 Factorial structures.- 3.8 Controlling for other variables.- 3.9 Intrinsically multivariate studies: two groups.- 3.10 Intrinsically multivariate studies: several groups.- 3.11 Further reading.- 4 Multivariate tests.- 4.1 Introduction.- 4.2 Towards multivariate tests.- 4.3 Multivariate test statistics.- 4.4 Choice of test statistic.- 4.5 Exploring noncentrality structures.- 4.6 Complications: the hypothesis matrix.- 4.7 Complications: the error matrix.- 4.8 Tests on multivariate analysis of variance assumptions.- 4.9 Tests on multivariate analysis of covariance assumptions.- 4.10 Further reading.- 5 Sequential and simultaneous test procedures.- 5.1 Resolution of general questions.- 5.2 Error inflation.- 5.3 Some basic ideas.- 5.4 Sequential tests.- 5.5 Gabriel’s method.- 5.6 McKay’s method.- 5.7 More on choice of test statistic.- 5.8 Variable selection.- 5.9 Confidence intervals.- 5.10 Further reading.- 2 The Practice.- Study A Alcohol relapse.- Study B Salsolinol excretion.- Study C Slimming clinics.- Study D Neuropeptides and psychiatric illness.- Study E Headache.- Study F Smoking styles.- Study G Alzheimer’s disease.- Study H Psychophysiological responses.- References.- Appendix Computer listings.- Listing 1 — SAS System A three group between subjects design (comprising one factor) with three variables in the within subjects design (unstructured).- Listing 2 — BMDP A three group between subjects design (comprising one factor) with three variables in the within subjects design (unstructured).- Listing 3 — BMDP A three group between subjects design (comprising one factor) with a three variable within subjects design (comprising one factor).- Listing 4 — SPSSX A three group between subjects design (comprising one factor) with three variables in the within subjects design (unstructured).- Listing 5 — SPSSX A three group between subjects design (comprising one factor) with a three variable within subjects design (comprising one factor).
